Language: English 切換至繁體中文首頁
NSB Oct 2026 Report 2.50M Disputed Posts 2.58M Daily GSN Attacks Voting Day: Nov 28, 2026

Taiwan 2026 Local Elections: Foreign Interference, Cyberattacks and Deepfakes — What Officials and Campaigns Have Reported

Ahead of Taiwan's nationwide Nine-in-One local elections on November 28, 2026, Taiwan's National Security Bureau, a mayoral campaign and politicians from both major parties have published figures, reported incidents or filed complaints regarding cyber operations, disputed online content, and artificial intelligence-generated media. This briefing compiles official disclosures from the National Security Bureau (NSB), cross-party complaints filed at prosecutors' offices, and statutory defense mechanisms under Taiwanese election law in a neutral, attributed reference format.

NPA Nov 4 Update Official Figures Reported Tactics Campaign Incidents Competing Perspectives Legal Safeguards Sources
Disputed Posts (Jan–Sep)
2.50 Million
Disputed messages tracked by NSB across platforms (+48% vs ~1.69M in Jan–Sep 2025)
Daily GSN Attacks
2.58 Million
Daily average intrusion attempts on Government Service Network through Sept 2026
July Attack Surge
185.6 Million
GSN attacks recorded in July 2026, exceeding the entire first quarter combined
Statutory Takedown
2 Days
Platform removal deadline under Article 51-3 upon police deepfake verification

Latest Update: National Police Agency Figures (November 4, 2026)

On November 4, 2026, the National Police Agency (NPA) held a press conference titled "Combatting Election Disinformation and Transnational Repression," hosted by NPA Deputy Director-General Kuo Shih-chieh (郭士傑) with case briefings delivered by Criminal Investigation Bureau (CIB) Commissioner Chiu Shao-chou (邱紹洲) (UDN, Nov 4, 2026; Taiwan News, Nov 4, 2026). According to the NPA, law enforcement agencies across Taiwan had accepted and investigated election disinformation cases through October 31, 2026 (CNA, Nov 4, 2026).

Disinformation Cases
184 Cases
Accepted by police through Oct 31, including 7 deepfake cases (UDN, Nov 4, 2026)
Referred to Prosecutors
112 Cases / 153 People
Transferred to judicial authorities for prosecution (CNA, Nov 4, 2026)
Top Online Channels
FB 46.4% / Threads 36%
LINE was 11.2%; nearly 70% spread via internet (UDN, Nov 4, 2026)
Transnational Repression
9 Probed / 4 Indicted
Suspected external force collusion with gangs and detectives (Taipei Times, Nov 5, 2026)

According to the Criminal Investigation Bureau, election disinformation has spanned text, image, and audiovisual materials, including AI-synthesized content, with nearly 70 percent transmitted online (UDN, Nov 4, 2026). Among the seven deepfake cases accepted and being investigated by police, reported incidents include a manipulated image titled "Wan-an Eliminates the Three Evils" (萬安除三害) targeting Taipei Mayor and Kuomintang (KMT) candidate Chiang Wan-an (蔣萬安), and a synthetic video falsely depicting Democratic Progressive Party (DPP) Changhua County magistrate candidate Chen Su-yueh (陳素月) saying things she never uttered (UDN, Nov 4, 2026). The CIB stated that, as of October 31, no candidate had yet applied for paid forensic examination under the Civil Servants Election and Recall Act (UDN, Nov 4, 2026).

The NPA noted that candidates who discover deepfake video or audio targeting themselves may apply to the CIB for forensic examination at their own expense from the election proclamation date of August 20 until November 27, the day before voting day, under the Civil Servants Election and Recall Act (CNA, Nov 4, 2026; Taiwan News, Nov 4, 2026). Candidates suspecting criminal violations may file a police report with the relevant recordings attached and submit takedown reports to online platforms (CNA, Nov 4, 2026). CIB Commissioner Chiu Shao-chou stated that using AI deepfakes to disseminate false information carries aggravated criminal penalties of up to seven years' imprisonment, according to FTV (FTV News, Nov 4, 2026).

The NPA reported that election-interference tactics used by external forces have continued to diversify ahead of the vote (CNA, Nov 4, 2026; Taipei Times, Nov 5, 2026):

  • Election betting manipulation: Manipulating illegal election betting pools (How Taiwan's election betting law works).
  • Sponsored trips to China: Sponsoring trips to China for Taiwanese citizens.
  • Donations via Taiwanese businesspeople: Channeling campaign donations through Taiwanese businesspeople, including hosting banquet events.
  • Temple and religious mobilization: Using temple and religious activities for organizational mobilization and soliciting local grassroots figures.

Regarding transnational repression, the NPA reported investigating nine cases suspected of involving external forces colluding with domestic and foreign gangs, unscrupulous private detective agencies, foreign individuals, and local collaborators (UDN, Nov 4, 2026; Taipei Times, Nov 5, 2026). Prosecutors have indicted defendants in four of these cases, while five remain under active investigation, according to FTV (FTV News, Nov 4, 2026). The four indicted cases announced by police comprise: the paint-splashing incident at Comedy Club (卡米地喜劇俱樂部遭潑漆案); the physical assault on Yaita Akio, CEO of the Indo-Pacific Strategic Think Tank (印太戰略智庫執行長矢板明夫遭毆打案); the covert trailing and filming of DPP Taipei mayoral candidate Puma Shen (民進黨台北市長候選人沈伯洋遭跟拍案); and the staged car accident causing injury to crosstalk performer Huang Yi-hao (相聲演員黃逸豪遭假車禍傷害案). According to the NPA Security Section, operational forms identified across these external-force cases include harassment such as paint-splashing or firecrackers, threats, violent assault or staged car accidents, and tailing or covert surveillance (UDN, Nov 4, 2026).

Note: InfoMiner Election Lab does not independently verify these National Police Agency statistics, reporting official law enforcement disclosures strictly as attributed data points.

1. Key Figures from the National Security Bureau's October 2026 Legislative Report

On October 20, 2026, the National Security Bureau (NSB) delivered a written operational report to the Legislative Yuan ahead of scheduled testimony by NSB Director-General Tsai Ming-yen (蔡明彥) before the Foreign Affairs and National Defense Committee on October 22 (CNA, Oct 20, 2026; Taipei Times, Oct 21, 2026). The document detailed network intrusion attempts and online controversial information collected across the first nine months of 2026:

Cyberattacks Government Service Network (GSN) Intrusions
  • Daily Intrusion Average: The NSB reported that through September 2026, the Government Service Network was targeted by an average of approximately 2.58 million intrusion attempts per day (CNA).
  • Upward Trajectory Since June: Intrusion frequency displayed an upward trend beginning in June 2026.
  • July Peak of 185.6 Million: In July 2026, detected cyberattacks surged past 100 million to reach 185.6 million, an amount exceeding the sum of all attacks recorded during the entire first quarter of 2026 (Taipei Times).
  • Suspected Preparatory Objective: The NSB said it cannot rule out that the rise is preparation (intelligence-gathering and intrusion) to interfere in the Nine-in-One elections (CNA).
Information Manipulation Disputed Social Media Posts
  • 2.50 Million Posts Tracked: By tracking abnormal accounts between January and September 2026, intelligence teams collected approximately 2.5 million posts containing controversial or disputed information (CNA).
  • 48% Year-on-Year Growth: This volume represented a 48 percent increase compared to the approximately 1.69 million posts collected during the identical period in 2025 (Taipei Times).
  • Platform Concentration: Posts were primarily observed across Facebook, online discussion forums, Threads, and YouTube.
  • Platform Shifts: The NSB noted significant proportional increases on Facebook (rising from 43.6% to 68.9% of tracked disputed posts) and Threads (increasing from 4.5% to 9%).
Indicator 2025 Baseline / Q1 Context Jan–Sep 2026 Reported Value Platform / Scope Details Attribution & Source
Disputed Online Posts 1.69 million posts (Jan–Sep 2025) 2.50 million posts (+48%) Facebook (68.9%), Threads (9%), Forums, YouTube NSB written report (CNA; Taipei Times)
Daily GSN Cyberattacks Upward trend since June 2026 2.58 million / day Taiwan Government Service Network (GSN) NSB written report (CNA)
July GSN Cyberattack Peak Exceeded total Q1 2026 cyberattacks 185.6 million attacks Surged past 100M threshold during July 2026 NSB written report (Taipei Times)

According to the NSB report (Taipei Times), AI is used across stages of attacks including probing for weaknesses, target seeking, searching for zero-day exploits and malware attacks via e-mail.

2. Interference Tactics Identified by the National Security Bureau

In its written submission to the legislature and subsequent testimony by Director-General Tsai Ming-yen on October 22, 2026, the NSB described specific operational methods, targeted domestic subjects, and delivery mechanisms the bureau attributes to China (CNA, Oct 20, 2026; Taipei Times, Oct 23, 2026):

Domestic Wedge Issues & Narrative Setting

The NSB reported that from January to September 2026, Chinese operations disseminated disputed information focused on domestic policy controversies—specifically energy, food safety, and national defense—with the suspected intent of disrupting government administration.

The NSB reported that operations also hyped specific candidates' policy proposals, public images, and the quality of local public infrastructure, aiming to shape local electoral outcomes.

Amplification Infrastructure & AI Integration

According to the NSB, narrative frameworks are frequently established by Chinese official bodies or state-run media, then amplified through proxy accounts, cyber armies (trolls), and commercial public relations agencies.

Tactics include content farms, clone websites, artificial intelligence-generated audio and video, fabricated social media accounts, and comment-section flooding. Operations monitor major news events, candidate platforms, election trends, and polling data to formulate cognitive warfare strategies, expanding the use of AI for information manipulation (CNA).

Targeted Demographic Groups & Financial Inducements

On October 22, 2026, NSB Director-General Tsai Ming-yen stated that election interference efforts in Taiwan are shifting toward specific demographic groups (Taipei Times, Oct 23, 2026). The bureau described several suspected practices:

Discounted Airline Tickets: The NSB said Beijing is suspected of mobilizing commercial airlines to provide discounted flight tickets for certain Taiwanese living in China to return home to vote.
Illicit Campaign Funding: The bureau stated it believes China is providing illicit funding to candidates and political groups it supports.
Cheap Tours & Suspected Bribery: The NSB cited suspected voter bribery carried out by offering cheap sightseeing tours to China.
Gambling Pools & Targeted Hacking: The bureau reported operations involving illegal betting pools, as well as hacking specific individuals to steal election-related intelligence for targeted interference.

Director-General Tsai stated that the NSB is utilizing artificial intelligence tools to trace the origins of disinformation, collecting evidentiary records of suspected interference, and hopes to hand materials over to judicial authorities before the November 28 elections.

3. Publicly Reported Campaign Incidents & Judicial Complaints

During the 2026 campaign cycle, a DPP campaign reported cyberattacks, and politicians from both the KMT and DPP have filed complaints or announced possible legal action over AI-generated media. These matters are listed below in chronological and attributed order:

DPP Campaign Incident Cyberattacks Reported by Puma Shen's Mayoral Campaign Taipei Mayoral Race

On October 29, 2026, NSB Deputy Director-General Chen Sung-chi (陳松吉) informed the Legislative Yuan's Foreign Affairs and National Defense Committee that the NSB is investigating suspected cyberattacks originating from China that targeted Democratic Progressive Party (DPP) Taipei mayoral candidate Puma Shen (沈伯洋) (Taipei Times, Oct 30, 2026; CNA, Oct 29, 2026). Chen stated that the investigation could lead to criminal charges against those found responsible.

According to Shen's campaign office and CNA reporting, several websites deployed by the campaign faced planned intrusions beginning on October 23, 2026, comprising two distinct attack incidents:

  • Host Server Permission Breach: In the first incident, according to the campaign office, attackers exploited a flaw in the website's upload function and gained administrative access to the host server, created fake administrator accounts and installed back doors (remote-control programs); they then tried and failed to obtain access to the main campaign website / internal system, because that system and its log-in credentials were not on the compromised server.
  • Traffic Flooding on Motorcade Route Sign-Up Platform: In the second incident, overseas IP addresses flooded a platform created for supporters to sign up for (claim) sections of the campaign's motorcade route ("洋流車掃認領路段網站"). Within an hour of launch, it received more than 70,000 fake sign-ups / authentication-code requests from overseas IP addresses, per the campaign; the campaign then required log-in to sign up, which halted the automated submissions.
Status (as reported): the NSB said its investigation is ongoing (Taipei Times, Oct 30, 2026). This page does not track later developments.
Cross-Party Legal Complaints Taipei District Prosecutors Office AI Deepfake Complaints Taipei & Tainan Figures

On August 3, 2026, Central News Agency reported that candidates and elected officials from both the DPP and KMT filed complaints against each other at the Taiwan Taipei District Prosecutors Office over suspected AI-generated audio (CNA, Aug 3, 2026):

  • Complaint by Chang Ming-yu (DPP): DPP Taipei City Councilor candidate Chang Ming-yu (張銘祐) filed a complaint alleging that Taipei Mayor Chiang Wan-an (蔣萬安, KMT) was involved in abetting the production of the video 「油不得你」 (a Taipei City Government public-health video on the carcinogen benzo[a]pyrene, according to the city spokesperson's description in CNA), which was suspected of using artificial intelligence to synthesize the voice of President Lai Ching-te (賴清德).
  • Counter-Complaint by Yang Chih-tou (KMT): KMT Taipei City Councilor Yang Chih-tou (楊植斗) responded by filing a criminal complaint against Tainan Mayor Huang Wei-che (黃偉哲, DPP) and DPP Legislator Wang Ting-yu (王定宇). Yang argued that Huang and Wang, during a press briefing, presented audio suspected of deepfaking the voice of Mayor Chiang Wan-an.
  • Law Enforcement Action: The Criminal Investigation Bureau received a complaint that the video was suspected of using AI to synthesize President Lai's voice, possibly constituting document forgery, and referred it to Taipei prosecutors. Prosecutors opened a case under 他字案 (a preliminary case docket).
  • Campaign & Official Positions: Mayor Chiang Wan-an previously stated that the allegations were intended to produce a chilling effect. A Taipei City Government spokesperson stated that artificial intelligence is a tool that should never be used for fraud or to harm people, but questioned what was wrong if it was used for health education videos on the carcinogen benzo[a]pyrene (BaP). Mayor Huang Wei-che and Legislator Wang Ting-yu stated that their public demonstration of audio files was aimed at alerting society to the risks AI deepfakes pose to personal character rights and information security.
Status (as reported): as of CNA's Aug 3, 2026 report, Taipei prosecutors had opened a case (他字案). This page does not track later developments.
Campaign Dispute New Taipei Mayoral Campaign Deepfake Image Dispute New Taipei Mayoral Race

On April 28, 2026, CNA reported an election dispute regarding alleged synthetic imagery in New Taipei City (CNA, Apr 28, 2026):

  • Lee Shu-chuan Campaign Statement: The campaign office of Kuomintang (KMT) New Taipei mayoral candidate Lee Shu-chuan (李四川) stated that an online illustrator had posted a deepfake image to smear him, adding that the office had gathered evidence and was planning to file a lawsuit.
  • Su Chiao-hui Office Response: The campaign office of Democratic Progressive Party (DPP) New Taipei mayoral candidate Su Chiao-hui (蘇巧慧) responded that the campaign team had never attacked opponents, had not issued press releases or held press conferences to do so, and had never authorized any internet users or social media fan pages to conduct electoral attacks.
Status (as reported): as of CNA's Apr 28, 2026 report, Lee's office said it was gathering evidence and considering legal action.

4. Debate Over the "Cognitive Warfare" Framing

Public discussions surrounding foreign interference and cognitive warfare have generated divergent perspectives between security officials and opposition spokespersons:

Security & Ruling Framework The National Security Bureau & Puma Shen

The NSB's report describes cognitive warfare operations it attributes to China: according to the NSB, these operations use proxy accounts, troll networks, content farms, and AI synthesis to amplify domestic wedge issues (such as energy, defense, and food safety); the NSB said the disputed information aimed to disrupt government administration (干擾政府施政) and influence the Nine-in-One elections (CNA, Oct 20, 2026).

DPP candidate and lawmaker Puma Shen has stated that cognitive warfare refers to conduct between states (UDN, Oct 1, 2026).

Opposition Critique KMT Spokesperson Liu Tsai-wei

On October 1, 2026, KMT spokesperson Liu Tsai-wei (柳采葳) publicly critiqued this characterization (UDN, Oct 1, 2026). Liu stated that Puma Shen uses academic definitions externally to pose as virtuous, but internally wields the national security label against domestic citizens, blaming the Chinese Communist Party whenever questioned.

Liu argued that this dynamic constitutes a partisan "green firewall" (綠色防火牆) used by the DPP to deflect domestic accountability. She stated that public grievances regarding domestic concerns—such as air pollution, vaccine shortages, and egg shortages—were labeled as Chinese narrative operations.

6. How InfoMiner Reads This: Scope & Methodology

InfoMiner Election Lab measures open-web public discussion volume and sentiment semantics; it does not attribute posts to foreign state actors and does not independently verify the NSB's figures.

InfoMiner Election Lab, operated by LargitData Ltd., monitors digital discourse using natural language processing (NLP) pipelines across public news websites, forums, and open social media spaces in Taiwan. Attribution of cyberattacks or covert influence operations to foreign intelligence agencies requires investigative powers and non-public data held by state security bodies such as the NSB.

Consequently, figures cited in this report—such as the 2.5 million disputed posts and 2.58 million daily GSN cyberattacks—are presented strictly as official government disclosures. InfoMiner tracks the public discussion footprint and sentiment surrounding these events without validating intelligence claims or taking partisan stances. Learn more about data collection and processing methods in the InfoMiner Methodology Whitepaper.

Frequently Asked Questions (FAQ)

How many disinformation and disputed posts were reported during the 2026 Taiwan election campaign?

According to the National Security Bureau (NSB) report submitted to the Legislative Yuan on October 20, 2026, intelligence teams collected approximately 2.5 million posts containing controversial or disputed information between January and September 2026 by tracking abnormal social media accounts. This marked a 48 percent increase compared to the 1.69 million posts collected during the same period in 2025. The NSB reported that Facebook and Threads accounted for 68.9 percent and 9 percent of these posts, respectively.

What did the NSB report regarding cyberattacks on Taiwan government networks ahead of the 2026 elections?

The NSB reported that through September 2026, Taiwan's Government Service Network (GSN) experienced an average of approximately 2.58 million cyberattack or intrusion attempts per day. Attack volumes began rising in June and reached 185.6 million attempts in July 2026—surpassing the combined total of all cyberattacks recorded during the first quarter of 2026. The NSB stated that it could not rule out that the increase reflected preparatory intelligence-gathering and intrusion efforts aimed at meddling in the November 28, 2026 Nine-in-One local elections.

How does Taiwan's Article 51-3 deepfake takedown law work during election campaigns?

Under Article 51-3 of the Civil Servants Election and Recall Act, from the formal election proclamation until the day before voting day, a candidate who identifies suspected deepfake audio or video of themselves broadcast on media or online may apply to police authorities for forensic examination. If police forensics confirm the media is a computer-synthesized deepfake, the candidate may issue a written request to the broadcaster or platform operator (copying the election commission). Broadcasters must stop broadcasting, and platforms must restrict access, remove or take down the content, within 2 days of receiving the request, retaining records for 6 months from receiving the request (extended to 3 months after a final judgment if litigation occurs).

What cyberattack and deepfake incidents were reported by candidate campaigns in 2026?

A DPP campaign reported cyberattacks, and politicians from both the KMT and DPP have filed complaints or announced possible legal action over AI-generated media. On October 29, 2026, the NSB confirmed it was investigating suspected cyberattacks originating from China that targeted DPP Taipei mayoral candidate Puma Shen's campaign websites, including a server intrusion in which, according to the campaign, back doors were installed, and a flood of over 70,000 authentication requests from overseas IPs. Separately, cross-party legal complaints were filed at the Taipei District Prosecutors Office in August 2026 involving DPP candidate Chang Ming-yu, KMT Mayor Chiang Wan-an, KMT Councilor Yang Chih-tou, DPP Mayor Huang Wei-che, and DPP Legislator Wang Ting-yu regarding alleged AI voice synthesis. In April 2026, KMT New Taipei candidate Lee Shu-chuan's office stated an online deepfake image had smeared him and prepared legal action, while DPP candidate Su Chiao-hui's office responded that it had never conducted or authorized any attacks.

Does InfoMiner verify the National Security Bureau's foreign interference statistics?

No. InfoMiner Election Lab measures public online discussion volume and sentiment semantics using automated natural language processing (NLP) pipelines across public websites, news reports, and forums. InfoMiner does not attribute online activity or cyberattacks to foreign state actors, nor does it independently verify the NSB's figures of 2.5 million disputed posts or 2.58 million daily GSN attacks. InfoMiner reports official disclosures strictly as attributed data points while analyzing how election issues resonate across open digital discourse.

What did Taiwan's National Police Agency report about election disinformation and deepfake cases in November 2026?

On November 4, 2026, Taiwan's National Police Agency reported that police had accepted 184 election-related disinformation cases through October 31, referring 112 cases and 153 individuals to prosecutors. Seven of these cases involved artificial intelligence deepfakes, including fabricated content targeting Taipei Mayor and KMT candidate Chiang Wan-an and Changhua County magistrate candidate Chen Su-yueh of the DPP, though no candidate had yet applied for paid statutory forensic examination. The Criminal Investigation Bureau found that nearly 70 percent of disinformation was circulated online, predominantly across Facebook (46.4 percent), Threads (36 percent), and LINE (11.2 percent). In addition, police investigated nine cases of suspected transnational repression involving external forces colluding with criminal gangs and private detectives, resulting in four prosecutorial indictments and five ongoing investigations.

7. Document & Reporting Sources